# final result: CHECK-1 failed - the invalid row was accepted before the check constraint was dropped.

### pre-drop insert result
# The test created the schema, table, and expression index, then attempted the invalid insert while the check constraint was still present.
CREATE SCHEMA
CREATE TABLE
CREATE INDEX
INSERT 0 1

### constraint and index readback
# The constraint catalog readback shows users_email_check still present and enforced before the drop.
   nspname    |      conname      | contype |                        definition
--------------+-------------------+---------+-----------------------------------------------------------
 check_case_1 | users_email_check | c       | users_email_check CHECK strpos("email", '@') > 1 ENFORCED
(1 row)

# The expression index remained present.
  schemaname  |       indexname       |                                       indexdef
--------------+-----------------------+---------------------------------------------------------------------------------------
 check_case_1 | users_lower_email_idx | CREATE INDEX users_lower_email_idx ON check_case_1.users USING btree ((lower(email)))
(1 row)

### drop and post-drop readback
ALTER TABLE
 conname
---------
(0 rows)

       indexname       |                                       indexdef
-----------------------+---------------------------------------------------------------------------------------
 users_lower_email_idx | CREATE INDEX users_lower_email_idx ON check_case_1.users USING btree ((lower(email)))
(1 row)

# The same invalid value was also accepted after the constraint was dropped.
INSERT 0 1
 id |     email
----+---------------
  2 | invalid_again
(1 row)

The key defect signal is the first `INSERT 0 1`: `invalid_again` was accepted before `users_email_check` was removed, despite the recorded predicate requiring `strpos(email, '@') > 1`. The successful drop and preserved `users_lower_email_idx` are separate post-drop results.